CH07 Chi-Square Test Medical Statistics医学统计学 演示教学.pptVIP

105 Table 7-12 inter-comparison of three methods Three、the comparison of many treatment group and one comparison group Example 7-10 We think the drug group is comparison ,physical group and plaster group are treatment groups , is there the total effective rate of two treatment groups and comparison difference? In the paired design experiment ,there are four possible results of the two treatments as to the each pair: ① positive number both of the two methods(a)； ② negative number both of the two methods(d)； ③ positive number of immunofluorescence, negative number of latex agglutination (b)； ④ positive number of latex agglutination, negative number of immunofluorescence (c)。 a, d are the agreement of the two methodsb, c are not agreement of the two methods Statistic: Cautions: The method is used for small sample Reasons : 1. only consider the ungreement condition(b,c) 2. not consider the sample size n and the conditions of the agreement(a,d) When the n ,a,d are large enough and the b,c are relative small ,there is nothing practical significance even if there is statistical significance. Steps of the test： Section 3 Fisher exact probabilities method in 2×2 table conditons： Basis of theroy：hypergeometric distribution not test Example 7-4， a doctor will study the precaution affect of the type B hepatitis immunoglobulin against intrauterine infection of foetus, and randomized 33 positive HBsAg patients into two groups：precaution group and nonprecuation group，looking at the table 7-4.Is there the difference between the two groups on the fetus infection ratio? table7-4 comparison between the two groups of fetus infection ratio of HBV 1.Basic idea: When the periphery total numbers of fourfold table are fixed, we can calculate the all combinations probabilities of the four actual frequencies, then make deduction according to the α level and the cumulative probabilities. 1．Calculate Pi : c